A revised multi-target detection approach in hyperspectral image

Constrained energy minimization (CEM) algorithm is very sensitive to noise and cannot classify the same type of targets with similar signatures, so we propose a revised multi-target detection approach, which is based on multiple-target constrained energy minimization (MTCEM), sum constrained energy minimization (SCEM) and winner-take-All constrained energy minimization (WTACEM). This approach revised autocorrelation matrix in original approach by excluding the desired target pixel vectors from the autocorrelation matrix. The simulation results indicate that the errors caused by the revised multi-target detection approach have decreased obviously.